The Beauty of Technology and the Pain of Industry: A Conversation Between Linear Capital and Portfolio Company Weijian Technology on 36Kr's "Undercurrent Waves" Podcast

Recently, Linear Capital partner Songyan Huang, along with Yanxue Liang — founder and CEO of Weijian Technology, a Linear Capital portfolio company — joined Waves Radio, a podcast produced by 36Kr's Dark Tides (暗涌Waves). Together with Guo Yunxiao, an analyst at Dark Tides, they explored how young people can understand traditional industries, starting from the theme of "the beauty of technology and the pain of industry." In this episode, Huang shared the story of investing in Weijian and his insights on backing frontier technology that lands in traditional sectors. Scan the QR code in the image below to listen. The following is an edited transcript of the podcast. Cover image generated by Tiamat.

Dark Tides: Our theme today is how young people can understand traditional industries, which connects to both of your backgrounds. Please introduce yourselves.

Yanxue Liang: I'm Yanxue Liang, founder of Shanghai Weijian Technology. After my PhD, I spent about ten years doing industrial robotics at FANUC in Japan. After returning to China and going through some choices, I settled on construction robotics. The company has been around for nearly three years now.

Songyan Huang: I'm Songyan Huang, partner at Linear Capital. I got my PhD from Zhejiang University in machine learning, then worked as a data scientist at Huawei. Later I switched to investing, focusing mainly on hard tech.

Dark Tides: How did you two meet?

Songyan Huang: I'll take this one — though "Professor Liang" (I'll call him that from here) probably still doesn't know how I found him. When I started in this business, I began with robotics investments. I mapped out all the Chinese researchers in robotics globally, especially the senior ones, and that's how I connected with him. Construction robotics was something he told me about; I hadn't even heard of it. After he explained it, I realized it was a massive market.

Yanxue Liang: I remember first meeting Songyan at a hotel restaurant. I'd just left Country Garden and was torn about whether to return to the new energy sector — I'd worked at CATL before. Songyan found me. At first we talked about industry in general, then a lot about construction. I couldn't tell what he was after. I wasn't thinking about starting a company then. After we connected, he'd reach out about once a month. By the time I decided to start up, I had some confidence because of that ongoing contact. In the end, it was Songyan's "egging on" plus my own restlessness that pushed me into it.

Dark Tides: Songyan, you suggested the theme for this episode — "the beauty of technology and the pain of industry." Why that?

Songyan Huang: To outsiders it might sound abstract, but the Weijian story makes it concrete. Professor Liang and his team have stellar backgrounds — top schools, top companies, excellent technical work. They could have lived very comfortably without starting a company. But if you've seen them in startup mode... I still have a photo on my phone of one co-founder who got fat from eating so many construction site boxed lunches — I tease him about it. He used to be very hipster. Another co-founder pushing a handcart on site. These images reflect the pain of industry. When you try to transform an industry, you're up against enormous inertia — production relationships formed over decades, centuries, even millennia. Can you get your hands dirty? That's the huge challenge for founders. These technical founders, we say they're holding the "dragon-slaying blade" — that's the beauty of technology. The two need to come together. So when I say "the beauty of technology and the pain of industry," it's an "and" — they need to match. The founder needs to dive in personally.

Dark Tides: What were your earliest impressions of traditional industries? Did you ever think you'd go so deep into one?

Yanxue Liang: "Traditional industry" has two parts: "traditional" and "industry." Industry implies scale — without scale, it's hard to call it an industry. "Traditional," in a stricter sense, means a production method that doesn't change dramatically over decades or centuries. Though after ten years, say, a car — replacing one every ten years felt fine before, maybe less so now. Phones, you replace every two years. But construction, agriculture — in China's current state, they probably haven't changed much in ten years. Though I do feel some loosening lately. So to me, "traditional" means both that the product form hasn't changed much, and that you don't sense changes in production methods or factors.

Songyan Huang: Professor Liang summarized it well. I'll just add: traditional industries change very slowly, and their production efficiency is relatively low.

Dark Tides: Linear has invested in many "technology + traditional industry" deals, many led by Songyan. Why did you spot so many opportunities there and act on them?

Songyan Huang: Linear gets excited about technology transforming industries, and our philosophy has always been tech for good. Beyond that, traditional industries face massive problems and challenges. Many are counterintuitive. Take construction — it accounts for over 20% of China's GDP. What supported that number? Migrant workers. When they age out, what happens? That's a massive problem, and it's already here.

Agriculture, livestock — same situation. Food security appears in the central government's No. 1 document every year. These are matters of national importance. But counterintuitively, or perhaps what people don't grasp, is that farming often relies on experience, or copying what others do. Yet these national priorities are so important — the problems within them need better technology and more advanced products to solve.

Dark Tides: Songyan, you've talked before about investment "going up to the mountains and down to the countryside" — visiting very grounded places: farms, chemical plants, cornfields. What did you see with your own eyes? You went to evaluate projects, but what else did you observe? Was it really that inefficient? Anything that shocked you?

Songyan Huang: Take pig farming. You might not have reached the farm yet, but within a kilometer, you'll definitely smell it. Once inside, you probably won't want to eat for the day. It's foul, but people have to work there. The staff are in there every day — you can imagine their working conditions, extremely poor. Farming — many farmers lose money growing crops. Some don't even own the land; they pay to lease it, and if the harvest fails, they lose even more. Construction — when we visit sites, we don't see young people. Almost everyone is 40 to 50, and from years of heavy labor, they look even older. That's the industry reality. From today's perspective, we still have time and a window of opportunity for technical people to transform it, or for the state to make adjustments.

Dark Tides: Next question for Professor Liang. You lived and studied abroad for a long time, in research institutes and corporate R&D. When you first thought of doing construction robotics and started engaging with the construction industry, what was your feeling? Did you encounter the kind of shocking scenes Songyan mentioned?

Yanxue Liang: In Japan I was in a research institute for about ten years. It was a bit of an ivory tower — no convenience store nearby, couldn't go anywhere without driving. That period gave me good intellectual training.

When you do technology long enough, you develop a certain way of thinking — like mathematics. You take that thinking and look at different industries. As mentioned, it's like holding a scalpel: you can operate on this industry or that one.

My transition from robotics to CATL's production lines was hugely formative for my thinking. Coming from a very deep, narrow specialty back to China, initially I knew nothing about building hundreds of meters of lithium battery production lines. But after three months, I started constantly summarizing the patterns of each piece of equipment — horizontal divisions, vertical divisions, and so on. Later I realized all equipment is fundamentally the same. That was a process of abstraction. Then taking that same thinking to look at construction or other industries — your whole perspective changes. What you know doesn't matter. The same thinking, the same logic applied to an industry is what counts.

Looking at the industry, you discover massive problems. There's a saying: a painting on the wall becomes background after three days unmoved. This explains why industry disruptors are always outsiders — only outsiders look at an industry with different eyes. If your thinking is rooted in lean principles, every industry looks problematic. Pick one and work on it. Since I came from robotics, and robotics is fundamentally about replacing one or more human motions — from that angle, looking at construction, a single motion like what we're doing with plastering robots: plastering is always just that one motion, yet that single motion represents a market of over 100 billion RMB in China. That's something people probably didn't see before.

Dark Tides: Have either of you had moments where you suddenly realized some very traditional thing was actually quite interesting, or any amusing stories from entrepreneurship or deal evaluation?

Yanxue Liang: The first time I connected an iPhone to a robot — we called it "zero downtime," a feature I worked on from 2013 to 2015. After completing it, I decided to return to China, because for the first time I felt the disruptive power of internet and cloud concepts on an entire industry.

This feature predicted how much longer a robot could work. Its significance: in automotive, one minute of robot downtime costs about $20,000. That's why domestic robots, if they don't reach certain stability levels, can't enter automotive production lines.

Once this feature worked, you could feel Industry 4.0 and its impact on the whole sector. That's when I decided to come back. I felt this shouldn't be limited to robotics — all industrial sectors could use it.

Songyan Huang: For me, through investing and engaging with different industries, I think more about problems. Here's an example: using AI for pig facial recognition — is this a real proposition?

First, it's completely bogus. The reasons are quite interesting. Visit a farm and you learn: one, when people pass through a gate, they'll cooperate and look at the camera. Pigs won't cooperate. Two, at a pig farm, pigs' faces are covered in mud — even if they cooperated, you couldn't recognize them. That's quite interesting. What it leads to: only by going into the industry do you know what's actually going on. What technology is feasible, what isn't. What looks feasible but is actually nonsense.

Dark Tides: In construction, especially the plastering process or segment, what have you observed?

Yanxue Liang: We keep talking about digitalization, but we never say the object — digitalization of what? Many industries conflate informatization with digitalization. Strictly speaking, some things should be called informatization. Digitalization corresponds to analog — can today's experience be expressed parametrically, written as code for automated expression, implemented through mechanization? The disruptive point in this whole process is whether experience can be parametrically expressed. If not, you don't need robots, because a robot is a digital device — it needs digital instructions to execute tasks.

So when we developed our plastering robot, I said first don't design the plastering mechanism. Bring in the master craftsman — every motion of his has no redundancy. We must explain what each motion means. When the master uses his hands, he's actually employing position control, force control, and visual control. But a robot hand has none of these — it only has a mechanical structure. Can we map these three dimensions onto one dimension? So step one isn't designing the robot; it's standing beside the master watching the true meaning of each motion, its scientific explanation. If we can explain it, then use a mechanism to implement it — only then can we digitize the most ancient traditional craft. Once digitized, you no longer need the master craftsman. The machine's self-evolution process has begun; subsequent evolution no longer relies on him. All China's plasterers will gradually face aging. Honestly, China's "infrastructure mania" has relied on the people's endurance and hard work. As workers age, this can't be sustained long-term. We must have alternatives. Plastering is one of the harder construction robotics processes. Going through this process, we learned how to use the same thinking to digitize everything around us.

Dark Tides: Our listeners are relatively young. Please help them understand more about plastering itself.

Yanxue Liang: When our houses are delivered as roughcasts, the interior walls you see have just been plastered. Because the walls aren't built perfectly straight, they must be smoothed — that's the plastering process. Smoothing relies on workers — generally one person can plaster 50 square meters a day, moving 1.5 tons of mortar for those 50 square meters. There's a vertical flatness requirement called "four vertical, four flat." And you have to press hard for adhesion — this is something you can't see or feel. When you knock on the wall and hear hollow sounds, the wall is already substandard. Our robot plasters with 30 kg of pressure. As long as your substrate and materials aren't specially problematic, there are basically no hollow spots.

Dark Tides: Songyan, regarding this plastering problem, have you seen others working on it?

Songyan Huang: About three years ago, construction robotics companies were quite rare. They've increased over these three years, but to date, only Weijian can do plastering — globally, no second company. Because plastering as a process is extremely difficult. The difficulties are numerous if we go into detail. The reality is: you'll find companies doing various other processes, but not plastering, because it's simply the hardest. From a technical angle. From an economic angle, plastering is probably the highest-paid among many construction processes. Combine these two: first, most expensive; second, no one doing it. Conversely, no one does it because it's hardest.

Dark Tides: Songyan, can you elaborate on what's happening in China's construction industry, or globally?

Songyan Huang: Globally, construction is a very heavy, very large industry — even more so for our country. Breaking it down, from property developers, government, down to the lowest tiers, even various suppliers like bathroom fixture suppliers — all belong to construction.

If we're talking changes, I see several points. First, automation or intelligence. But why now? The industry used to be very profitable — many fortunes came from it. When an industry makes that much money, people don't think about other issues. And there was the older generation of migrant workers. From production factors to industry economics, no one considered reform.

Today, because the industry itself is changing — it's not as profitable as before — these factors start driving it toward cost reduction and efficiency gains. China's aging population is also a factor. These forces push the industry to change: efficiency improvements, productivity changes in various processes, starting to be applied and spread within the industry.

Dark Tides: Many traditional industries are changing, a process driving cost-reducing, efficiency-improving technologies to replace past inefficiencies — opportunities emerging within this.

Songyan Huang: Though we can't generalize. Construction is driven by what I just described. Farming is a different issue. People say America's mechanization and digitalization are excellent. What about China? Why not as good? China hasn't even achieved intensification yet. When families each had their small plot, you couldn't provide services — commercially unviable. Only when an industry moves toward intensification do you get the chance to accomplish automation and digitalization, to introduce more efficient productivity tools. Different industries differ — their trends differ, their driving factors differ.

Dark Tides: In recent years, some tracks with traditional industry elements or attributes have returned to investors', entrepreneurs', and ordinary people's视野. New energy vehicles might be a good example — both the long chain of traditional auto industry and very fresh technology. Why now, and what experiences from this traditional industry, especially operators' or investors', can apply to how we view these traditional industries now?

Yanxue Liang: In traditional industries, analyzing what problems exist — it's hard to say whether it's profitable, what opportunities exist, whether current operations are reasonable. But if you divide a person's activities into value-adding and non-value-adding, analysis becomes easy. At this point, can new technological elements bring change to this industry? That's mainly my angle. Because traditional industries — take hairdressing, very traditional, but hard to change.

Looking at construction today, you find its production factors are labor, plus project managers, migrant workers — these factors can potentially change. Physical replacement is labor factor replacement, one approach. Another is whether processes are reasonable — if unreasonable, they can be reshaped. From this angle, first, this industry's production factors can be upgraded; the current state can't be maintained. With aging and labor reduction, we already know clearly: today's 100 yuan won't become 110 tomorrow, it'll be 99 — a declining trend. Aging, fewer people, yet scale increasing — this contradiction can't be reconciled.

Production factors already determine that production efficiency will decline. At this point, some means must replace them — new technology, new methods, or some underlying logic. When you find possible relevant technologies to solve it, this approach applied to any industry, I think, can produce similar results.

Dark Tides: Weijian is still a startup — startups may have many immature aspects. Is something this traditional suitable for startups to tackle? What's your view?

Yanxue Liang: Depends what's needed. Pure technology, or way of thinking? Technology matters in entrepreneurship, but technology itself doesn't create economic value — you must make it into a product, a commodity. Making commodities in some big-B industries, especially traditional ones, requires understanding them and cooperating with them. In China's current state, regardless of industry, leaders are very pragmatic. If you can explain things clearly, they're willing to try. As we are now — the original production method is set, suddenly a robot enters, and the social division of labor in industry versus construction are completely different. Bringing an industrial machine into construction necessarily creates major shock to existing organizational structures. But we've found some central SOE leaders genuinely very open about this. They're willing to compromise, but the premise is you must show them your potential — you must vividly demonstrate core value before them. If my robot is 3x human efficiency, they might not feel much. But if we can do 8x, they'll immediately feel they can compromise for you — even with discounting, it's still 3-5x. Ultimately, it depends whether your approach can convince the other party's leader from underlying logic. Honestly, I feel transforming traditional industries must be done by startups — change always happens at the edges, where the industry's center can't reach. The center is impossible. That's my feeling.

Dark Tides: Your fields may be much harder tech. Songyan, from an investor's view, how do you see such startup projects? What are common difficulties, or advantages?

Songyan Huang: First, you must dive in personally. If you don't, investors usually ask founders: "What's your moat?" This is unanswerable. I often tell our team: don't ask these questions directly — they're invalid.

First, figure out the problem yourself. Talking technology without the problem is invalid. For both investors and founders, diving in tests whether the founder can abstract the problem — from various facts and matters, can you abstract a problem? With the problem, you can think how to solve it, then what technology to use. Technology combined becomes a product; then how to commercialize it into a commodity.

After this whole chain, it's very clear. If you can ask these questions clearly, you clearly see whether the founder has thought it through. If they haven't even defined the problem, it's obvious in discussion — and you naturally know who can and who can't.

Similarly, newcomers to another traditional industry can borrow this thinking chain. If you can think through this chain, you'll naturally understand that industry.

Dark Tides: Since we're on understanding an industry — many listeners are young, some very young aspiring entrepreneurs, or in startups, or young investors. Songyan and Professor Liang, can you summarize some experience on how to learn a relatively traditional industry? After diving in, what to look at?

Yanxue Liang: If looking at an industry with a methodology — I early on, and still do, looked from a robotics angle, which may not apply to others. Robotics' core factor is the physical robot's actuator replacing one or more human motions. As mentioned, focus on which changing factors bring value-add — wherever there's value-add, there's natural barrier, it's hard to do. Why construction sites distinguish small workers from master craftsmen? Masters need skill; small workers just move materials — you can do it, but it doesn't pay. If you do master work, as long as you succeed it's fine, because there's natural barrier.

For entrepreneurship, I think it's about tackling hard problems — doing something where, once you do it well, others can't easily enter and compete. China has too many entrepreneurs; if you choose somewhere without barrier, either it doesn't pay, or it's easily copied — you struggle just to get by. That's a personal choice.

So I can only say I look from a robotics angle, accustomed to this way. The analysis method: how many degrees of freedom does this motion need? How much value-add? What's the market size for this motion itself in China, globally? For other aspects I may not have much advice.

Songyan Huang: Advice for new investors. First, traditional industries aren't as nice as imagined — production relationships formed over decades or centuries, transforming them is extremely difficult. So when evaluating a company or product, the key is whether there's opportunity to bring order-of-magnitude efficiency gains at certain points. Why could original production relationships persist so long? Because of replacement costs. The value from efficiency gains, combined with actual circumstances, needs discounting minus replacement cost. This you need to see. How to calculate this clearly requires very deep industry understanding. So on one hand, very strong cognition of industry problems; on the other, very deep, detailed understanding of technology.

About Linear Capital

Linear Capital is an early-stage investment firm focused on "frontier technology + industry," covering frontier technologies such as data intelligence, digital new infrastructure, next-generation robotics, and new technological transformations in traditional domains (biomedicine, materials, energy, etc.), applied across vertical industries to substantially improve efficiency, empower solutions to pain points, and complete industrial upgrading — achieving excess returns through substantial increases in industrial value. The firm currently manages ten funds with total AUM of approximately $2 billion.

Our investment stage focuses primarily on leading angel to Series A rounds, with typical check sizes of $3–8 million or RMB equivalent.

To date, Linear has made early investments in over 120 startup teams including Horizon Robotics, Kujiale, Sensors Data, Tezign, Rokid, Guandata, Agile Robots, and others. The combined valuation of Linear's portfolio companies is approximately $20 billion.

In the near term, Linear Capital is working to become the premier "Data Intelligence Technology Fund"; in the long term, it aims to build the most influential "Frontier Technology Application Fund."